What are the additional factors or variables, if any, that should be considered to enhance the robustness and generalizability of the calculators?

From the specific recommendations:
Distribution Assumptions:

Variability Consideration:

Incorporation of Simulations:

Calculation of Additional Parameters:

Expanded Range of Default Values:

Flexibility in Output Selection:

Consideration of Time-Related Factors:

Inclusion of Pilot Studies and Simulations:

Sampling Strategy Optimization:

Focus on Effect Size Over Significance:

An ability to use a different sample (from a pilot or another study)  to estimate exposure  or biomarker.
Addition of a parameter that indicates how far back the exposure of interest is (number of year) as compared to when biomarker  was obtained.

Please see the above response. 

A brief explanation of how to convert conventional input variables to standardized units.
I don't see the utility of Section 2.  The "motivating examples" are separated from the corresponding "data," and no data are shown.

I can't think of any other factors or variables to include.

The comments below actually pertain more to the presentation of this white paper than specific factors or variables.

GENERAL COMMENTS

 

How will this white paper be released?  As a publication in a scientific journal?  As a stand-alone tool on the web, together with links to the calculators?  This question is relevant because I suspect that the typical structure for scientific papers (Introduction, Methods, Results, Discussion) may not work as well for the purpose of this paper.  If the authors have freedom to arrange the paper as they like, I suggest that they consider something more like the following:

Expanding on the last point, how do these calculators advance the field?  Have the authors done their homework to see what is currently available?  If there is nothing available that addresses these study design questions, then state it.  If there is something available, it would help to compare the calculators presented in this paper to what currently exists, and argue why the presented calculators are better.  For example, a VERY cursory online survey of currently available sample size calculators resulted in the following (many of which ignore within-person variability):

 

 

1. INTRODUCTION

 

Sections 1.2, 1.3, and 1.4 all seem to say similar things; this seems redundant and could be made more concise.  As a specific example, all background and description of ICC could be combined and presented in one place near the beginning.

Section 2.3.  Example 3: Longitudinal Studies on Chronic Disease Progression.  “The ICC becomes particularly relevant in these studies, as it reflects the consistency of biomarker measurements within individuals across multiple time points, if we assume that there is no time-trend in the studied exposure.”  As written, this seems odd – it is precisely the within-individual variation and the time trend that is of interest to study in chronic disease progression, so the between-individual variation and hence the ICC is of lower concern.

SECTION 2.  MOTIVATING EXAMPLES AND DATA

What is the benefit of presenting Section 2 – Motivating Examples and Data?  It doesn’t seem to relate to the rest of the paper.  If the authors used those data to illustrate the calculators they present, that should be explicitly stated.  Otherwise, there seems to be little reason to include this section at all; I suggest deleting it. 

 

If the authors feel strongly about including Section 2, perhaps compress it to a short paragraph placed in the Introduction.  Alternatively, if the authors truly want to keep it in the paper at this level of detail, this section would be simpler to follow if it was rearranged, such that each background example (e.g., section 2.1, Example 1, Assessing the Impact of Air Pollution on Respiratory Health) was immediately followed by the data example (e.g. section 2.5, Data Example 1: EPA Air Quality Monitoring Data).  

SECTION 3.  METHODS

Are the parameters for the examples in this section drawn from actual data?  (E.g. EPA data on air pollution)?  If so, it would be helpful to state that clearly for each example.  Otherwise, readers could assume data are all hypothetical and thus perhaps of less relevance.

Sections 3 and 4.  It might help to add a paragraph about how the calculators presented in this paper relate to each other, or a suggested order for use by a researcher designing a study from scratch. 

SECTION 4.  RESULTS

If the authors are able to arrange the paper as they like, I recommend rearranging the Methods and Results sections to put the Results immediately after the corresponding Methods.  Example:  Start with 3.1 – Calculator #1: The Number of Repeats Calculator, followed immediately by what is currently in 4.1: Number of Repeats Needed for Desired Validity Coefficient.  I would move Figure 2 immediately after 4.1, so the reader doesn’t need to flip back and forth between the Figure and the text that describes it.  

Many if not most of the Results are intuitively obvious and do not add to the literature; some readers may wonder why their time is being wasted.  Example in 4.1: “The number of repeats needed decrease as the ICC increases” or in 4.2: “When we need more precise estimates and exposure is more variable one can anticipate more resource-intensive research, far more so than if there was not intrinsic variability over time within a person.”  However, this can be acknowledged and used to bolster the validity and hence usefulness of the calculators, by adding something like the following sentence when the first results are presented: “The calculators yielded results that confirm what we would expect, indicating validity and utility in the field.  Several specific results are listed below…” 

SECTION 5.  DISCUSSION

Neither section 5.1 or 5.2 seem to add anything really new to the literature.  I suggest combining and compressing them.

What would really help is a discussion of how these calculators compare with already existing calculators, and thereby add to what is already known or available.

In the Discussion, it could be helpful to expand on practical aspects of using these tools, considering additional approaches in study design.  For example:

Some of these tools require inputting parameter values that researchers might not be sure of for the compound(s) of interest, or that might not be readily available.  Some potential solutions you might suggest in this white paper:

Any models with time (longitudinal data, time to event,). 
e.g. cox models are very common in my field. 

The ability to include additional independent variables (that are correlated with the exposure variable of interest) in the calculators would reflect real word exposure scenarios more appropriately.